Chad is an AI assistant that answers questions about United States political financial disclosures by querying a structured database of them directly. It reads congressional Periodic Transaction Reports filed under the STOCK Act and executive-branch OGE Form 278 filings, and answers in plain language with the underlying rows attached. It is a research tool for reading filings that have already been made public. It does not offer investment advice and makes no claim about what any security will do next.
The database holds 157,416 disclosed transactions from 526 filers, covering transactions dated January 3, 2011 through August 7, 2026, alongside 13,332 individual holdings lines extracted from 179 executive-branch OGE filings by 55 officials. The most recent disclosure on file is dated September 2, 2026.
A general-purpose model can describe what a Periodic Transaction Report is. It cannot tell you the median disclosure lag across 157,416 filed transactions, or which of a committee’s reconstructed positions are held by members of both parties. Those are queries against filed rows, not recall. Every figure below is computed from the database described above.
Across the database there are 31,501 transactions by members of a committee with oversight of the traded security’s sector, dated within 30 days before a scheduled hearing or markup of that committee. 17,089 of those are purchases. The median gap between the transaction and the event is 3 days, and the median gap between the transaction and the date it became public is 17 days — meaning the trade was typically disclosed 14 days after the hearing it preceded.
Proximity is not evidence of wrongdoing, and this figure counts co-occurrence rather than causation. Committee members are frequently invested in the sectors they oversee for reasons unrelated to any specific hearing.
Chad reads the ethics-agreement rows out of a nominee’s filing, then looks for the named assets in that official’s subsequent filings and transaction reports. The database holds 217 ethics-agreement commitments across 49 officials, and 13,332 holdings lines to check them against.
Stated limit: we do not publish a resolved-versus-still -held scorecard here, because the agreement text is free-form prose rather than a structured list of assets and deadlines, and matching it to later holdings is currently heuristic. Publishing a divestiture verdict about a named official on that basis would not meet the standard of the rest of this page. The per-official comparison is available interactively, where the underlying filing rows are shown alongside it and can be checked.
1,305 disclosed transactions are options rather than shares. Of those, 1,081 have a strike price and 1,009 an expiration date, both extracted from the asset-description text on the filed form. 694 are calls and 611 are puts.
Where the underlying’s price on the transaction date is known, Chad classifies the contract against its strike: 204 contracts currently carry that classification — 109 out of the money, 63 in the money and 32 at the money at the time of the disclosed transaction. The remaining 1,101 are unclassified because the strike, the expiry or a price for that date is missing.
Chad stores returns from two different starting points. One measures the 30 days after the transaction date; the other measures the 30 days after the disclosure date — the first day the information was public. Across the 92,616 transactions where both are populated, the median 30-day return from the transaction date is 0.63% and the median 30-day return from the disclosure date is 0.90%.
Those two medians are close, and the disclosure-date figure is the slightly higher of the two. On this measure the filings do not show a typical timing advantage that decays by the time the public sees them. We report medians rather than averages because a small number of rows carry corrupted price data that would distort a mean.
The median disclosure lag is 20 days. 11,054 transactions were disclosed more than 45 days after the transaction date, which is the STOCK Act reporting deadline.
Chad reconstructs a running position book for each member from their buy and sell filings, then aggregates by committee. For House Financial Services, the reconstruction covers 55 members with disclosed transactions out of 132 who have served on the roster, spanning 8,773 transactions across 452 distinct securities. The most widely held position is NVDA, disclosed by 8 members. 69 of the committee’s positions are bipartisan — held by at least one member of each party.
Chad matches transactions to bills by the filer, the bill’s policy area and the affected sector, within a 90-day window either side, and reports the gap in days. There are 231,667 such matches against bills the filer sponsored, involving 9,515 distinct bills and 310 filers. The median gap between the bill’s introduction and the transaction is 3 days.
A single transaction can match several bills, so the match count is much larger than the number of transactions involved. Treat it as a starting point for looking at specific pairs, not as a count of distinct events.
Chad extracts every holdings line from every OGE 278 filing it has, keyed to the filer and the filing year, so a nominee-year filing and a later annual filing can be set side by side. Across 179 filings from 55 officials it holds 13,332 holdings lines, of which 8,544 carry a reported value range and 1,999 resolve to an individual listed security.
The gap between those last two figures is the honest shape of this data: most lines on a 278 are funds, partnerships, trusts, cash accounts or municipal bonds rather than individually listed stocks, so they carry no ticker to resolve. For an official with both a nominee filing and a later annual filing, Chad shows which lines appear in both, which disappeared and which are new.
Chad stores, for each transaction, the number of days to the underlying’s next earnings date and what that report did. 6,975 transactions fall within 14 days before an earnings date. Where the subsequent report is known, 3,639 preceded a positive EPS surprise and 988 preceded a negative one. Most transactions in the window have no known subsequent report attached, so those two counts do not sum to the total.
Chad flags a transaction as clustered when several filers disclose the same security within a short window. 1,996 transactions carry a cluster flag, forming 476 distinct clusters, of which 107 are in an individually named company rather than a pooled category such as municipal bonds. The largest cluster in a named company involves 6 filers disclosing SpaceX in June 2026.
157,416 transactions from House and Senate filers, sourced from the Clerk of the House and the Senate Office of Public Records. Each row carries the filer, the security, the transaction type, the reported amount range, the transaction date, the disclosure date, the owner (filer, spouse or dependent child) and a link to the source document.
179 filings from 55 officials — Cabinet secretaries, agency heads, White House staff, the President and Vice President — covering nominee, new-entrant, annual and termination reports. These yield 13,332 holdings lines plus ethics agreements, liabilities, gifts and outside compensation.
Filings are ingested nightly. The most recent disclosure in the database is dated September 2, 2026. Because filers have up to 45 days to report, the most recent transaction dates typically run several weeks behind the most recent disclosure dates.
Amounts are ranges, not exact figures. Congressional filers report in brackets — “$1,001 – $15,000” and so on — so any total built from them is an estimate bounded by those brackets, never a precise portfolio value.
A disclosure is not a trade you can follow. The median disclosure lag of 20 days means the information is weeks old when it becomes public, and as the two medians above show, the 30-day return measured from the disclosure date is not systematically worse than from the transaction date.
Not all filings name a security. Many lines describe funds, partnerships, trusts or municipal bonds with no listed ticker. These are retained with a category label rather than discarded, which is why counts of “transactions” and counts of “listed securities” differ.
Nothing here is investment advice, and none of it is a statement about what any security will do. Every figure describes filings that have already been disclosed.
A free account includes a monthly allowance of questions. Paid plans raise the allowance and unlock longer analyses. See plans and pricing.
Yes. The same database is available as a REST API with webhooks that fire when new filings are scraped. Commercial use of the data is permitted on self-serve plans. See the API access page and the developer documentation.
Public filings: the Clerk of the House, the Senate Office of Public Records, and the U.S. Office of Government Ethics. Chad links to the source document for the rows it cites, so any answer can be checked against the original filing.
Yes. Chad reads filings that are themselves handwritten or scanned in many cases, and extraction from them is imperfect. Answers carry the rows they were derived from precisely so the answer can be checked rather than trusted. Where a figure on this page is uncertain or partial, it is stated as such above rather than rounded into a cleaner claim.
Figures on this page are computed from disclosed filings. Transaction counts and the latest disclosure date are current as of September 2, 2026; the remaining figures were derived on August 13, 2026.